problem with the calculator in colorsync

I discovered the calculator "panel" in Colorsync, which is able to show the color of the pixel under the pointer. By default, it shows RGB color in "floating point". (Comment: RGB should default to 8-bit).


Part 1. It is fairly easy to mess up the interface of the calculator so that the magnifying glass sliders work but the numbers do not. For example, select RGB in the left panel and HSV in the right. This creates the effect. It also reverts the units to floating point while greying out the box so that you can't change "it" back to RGB. The problem is that RGB and HSV or Lab need different units and the interface doesn't allow for that. Also, the dropdown menu where it offers a bunch of profiles to choose from shouldn't be there or should contain only profiles that are relevant -- it just reverts to sRGB after you choose something else. If you choose RGB to Lab color and drag the RGB sliders around, the Lab values are peculiar 4 digit numbers. This interface needs some work.


Part 2. I was using it to try to understand color as shown in Photoshop, using a test file constructed with particular values. I was not happy to see that the RGB number reported for the patches differed between the colorsync display and the Photoshop report of the color of the same block of pixels. After more tinkering around, it appears that Photoshop changes the displayed effect of the blocks of pixels to be more visible or something.

For one grey scale patch:

file value: 24 24 24

Colorsync: 26 26 26

Photoshop: 37 37 37 (in Adobe RGB)

Photoshop: 32 32 32 (after converting to sRGB)


For a strongly colored patch:

Colorsync: 250 23 135

Photoshop: 255 1 131


This is with the standard iMac profile (5K). The numbers change somewhat with a different profile but the overall relative differences are still there. I don't know how to think about this when it comes to color and color management. As if this topic wasn't complicated enough already.


The documentation talks about the calculator as a color space converter for pixels on the screen or values entered by hand, not mentioning the confusing issues.

Posted on Apr 8, 2015 12:59 PM

Reply
11 replies

Apr 10, 2015 5:30 PM in response to Kurt Lang

I'd like to follow up on this because I don't think I quite understand the issues so maybe you can explain.


Photoshop. The working color space and the color space of the file are sRGB. The proofing is set to "Monitor RGB" which i believe to be the system's standard icc profile for the display, in this case iMac. "Proof colors" is on. I thought this would mean Photoshop showing the file with the Mac's own monitor colors. Please correct misunderstandings.


Preview. The display prefs are set to iMac. I understand that Preview is "color managed".


So here I have two apps that are showing the same file. Unless I have got it way off base, both apps are showing the file under the same color assumptions and I could expect the measured RGB values to be the same, at least relatively speaking.


The next step would be to measure the RGB values of the on-screen pixels using two different tools, Photoshop or ColorSync calculator. But if I'm wrong so far, there is no point continuing, so please let me know.


My goal here is to try to understand something about what I'm seeing on the screen and what icc display profiles actually do. And coincidentally whether the ColorSync Calculator is any use in this process.


Thanks.

Apr 14, 2015 12:56 PM in response to Jan Walker

Hello Jan,


There's no need to have the proofing choice on in Photoshop for your monitor's color since it will do that anyway. No matter what color space an image is in, Photoshop first displays it in its interface according to the file's color space. Since that's meaningless to your monitor, it then translates the color to your monitor's color space in the background. That's the color you see. It's always your monitor profile as the color Photoshop actually displays.


I created a very simple TIFF image of block colors. TIFF since color will remain completely flat. A JPEG, even at the highest quality level will introduce pixel values that aren't the same as the value you chose for a flat color.


Knowing that, Preview once again exhibits its complete uselessness as an image editor. The only thing I ever use Preview for is what its name implies; a viewer. Nothing else. So what's wrong with it? Well, a lot.


This is the test strip I made of highly saturated colors. The top using my native monitor's color space as the working space. It's somewhat larger than Adobe RGB in some colors, a bit less in others. The lower is the same strip converted to sRGB. Since it's a much smaller color space than my monitor, most of the colors get noticeably knocked down to a less vibrant color, which is the nearest color to the original sRGB can handle.


How well you can see the difference depends on your monitor's capabilities.


User uploaded file


Now, here's the same two images as Preview in Yosemite displays them.


User uploaded file


What's with the banding in the cyan? It sure the heck isn't in the file. Only about a third of it displays as the correct color. In the sRGB image, it's split in the same area, but displays slightly lighter rather than darker. The green on the top also impossibly displays slightly lighter than sRGB.


There really isn't a way to know what Preview is doing as far as color management. There are no settings you can choose for display, only assigning a profile to an open image. Which is something you normally only do to an image that has no embedded profile.


Then it gets worse. The color readings from the system tools don't agree with each other. Not even the color it displays!


User uploaded file


What is the above supposed to tell me? At least the Digital Color Meter is showing me the same hue and saturation as the magenta patch I had the cursor over. The ColorSync Calculator is showing it as a purple; even on the right which is my monitor profile. I know it always gives decimal values, but what use is that? Nobody reads RGB values that way. It's meaningless to most people to give the values as a decimal.


Then Preview really gets useless. I opened the same image in Photoshop and Preview. Both Photoshop and the Digital Color Meter show the correct RGB value of the green patch as 43, 233, 38. As soon as I move the cursor down over Preview's display, the Digital Color Meter changes to 123, 238, 39.


User uploaded file


In Preview's defense (I suppose), it's passing the green value it's displaying to the Digital Color Meter. Which would be the proper values if it were displaying the image correctly to start with; which it's not.


So basically, ignore Preview for any type of image editing. Don't even know what to tell you about ColorSync's Calculator function since it is also clearly wrong when it won't display magenta as magenta. It was also impossible to get the Meter to show me any calculations in the order I want. If I set the left menu to my monitor's color space, Adobe RGB or anything else, the instant I clicked the magnifier to get color readings, it changed back to sRGB! So no matter what you do, you always get sRGB color values as the source color, whether you want them or not. The conversion then to the color space you choose on the right is always meaningless when you have no desire to start with sRGB readings.

Apr 12, 2015 6:37 PM in response to Kurt Lang

Thank you very much for this information. And for answering the question about whether ColorSync Calculator is useful for anything. It isn't. Whatever is it doing in the shipping software I wonder.


Oddly enough, on my iMac 5K screen, looking at this post with Safari, the color patches in your examples look almost exactly the same in their two versions. But I do know what you are describing because I see those peculiarities when I print out test images that have a full range of colors.


Thank you for pointing out the digital color meter. This might be what I'm looking for, which was a way to examine "color" on my monitor as I change from one profile to another. And to examine how Photoshop displays colors - with some kind of curve e.g. a grey patch with RGB value of 24 in the file is shown as 32 on the screen.


I see the color meter reports sometimes RGB with the notation "clipped" usually when the channel is 0 or 255, but not every time (see example below).


Now for the next question. The drop down menu in Digital Color Meter offers several color spaces, including the one you used, "Display native values". What the heck does that mean? It makes a huge difference on my screen in the values reported for "display native values" and "display in sRGB". For example, the cyan patches in your first pairing:

Native: (top) 0 242 228 and (lower) 58 253 244

sRGB: (top) 0 (clipped) 244 230 and (lower) 0 255 246


Native to what?


So with the sRGB report, the R=0 is described as clipped in one patch but not in the other. Is the color meter just not reporting right? or is the second set not actually clipped but just sitting on the boundary of the sRGB space.


Many thanks!

Apr 13, 2015 11:50 AM in response to Jan Walker

Hello Jan,


There isn't a short way to explain this, so please bear with me.


I figured out what Preview was doing in the above where the cyan patch was split, and everything to the right of that displaying incorrectly. The patches up to the band were the correct color, and the rest of the image was displaying as sRGB. On the sRGB version of the strip, Preview was correctly displaying sRGB values up to the same point, but the rest of it to the right - well - who knows what that was supposed to represent. So there's one bug to submit to Apple.


Second bug is ColorSync's Calculator. It's just kind of backwards how you use it. You pick the color with the magnifier you want a reading of first, then choose the color space you want from the menus. The color shown is still wrong, though. If I select the same pink, and then set the drop down menu to my monitor profile (which is the color space the strip of patch colors is in), they should perfectly match. You can see below they don't. The small rectangle inside the pink area is how the calculator is showing what is supposed to be exactly the same color. The Digital Color Meter even reads them as different RGB values. So bug number two to submit.


User uploaded file


I see the color meter reports sometimes RGB with the notation "clipped" usually when the channel is 0 or 255, but not every time (see example below).

That's actually correct. It would be clipped in sRGB. The two main things to know about color as a computer understands it (and any device you can use with a computer that handles color, such as a digital camera, scanner, printer, etc.), are these:


1) The only color space in which each and every color is a fixed point, and will always be that color no matter how many times you express a given value, is L*a*b*. CIE L*a*b* is the current model we use. As a short explanation, it's every color we are capable of seeing defined as a mathematical model so computers have a consistent reference point. Not that the same L*a*b* color will look the same on every computer since they are all over the place as far as the color range a particular monitor can display, and how it's been profiled.


2) All RGB color spaces are a portion of L*a*b*. It tells the computer how much of L*a*b* a particular scanner, monitor, printer or whatever is capable of reproducing. Everybody makes a big deal out of Adobe RGB being some sort of standard. Well, I hate to be the one to break it to everyone else, but it's just one of millions of RGB color spaces. All of which are just as "independent" as Adobe RGB (2008) is claimed to be. If I took my monitor profile, which outdoes Adobe RGB's range in many hues, and called it Adobe RGB (2015), no one would known the difference. At least until Adobe made it known they didn't release such a color space. But it would be just as valid at being an independent color space as any other RGB profile.


So, what is clipping? Take a look at this comparison:


User uploaded file


This is Adobe RGB and sRGB displayed together. Adobe RGB being the larger space. The points circled in red are both the most saturated green each space can display. Both of which will read 0,255,0 when working in that color space. The points circled in green are those two space's most saturated red (which don't look very red due to the angle of the 3-D view). Both of which will read as 255,0,0. The next picture will make this easier to visualize:


User uploaded file


The patch on the left is the pure red on my EIZO monitor. The one on the right is the brightest red, also defined as 255,0,0, that sRGB can show. Obviously note even close to the same color, despite both being 255,0,0. The giveaway why they're different are the L*a*b* values for each one. They are at completely different L*a*b* coordinates, just like the green circled areas above.


That's what clipping means. sRGB is incapable of having the native hue exist within its space, so it lets you know that you are working with a color that is outside of sRGB's realm. And just for fun, let's convert the 255,0,0 sRGB color pictured just above to my monitor profile. Look what happens to the RGB values:


User uploaded file


It's not 255,0,0 anymore. In order to keep that color in exactly its same point of reference to L*a*b*, it became 206,43,13 in my monitor's RGB profile space. If I were to do it the other way around (my monitor's 255,0,0 color to sRGB), then it would still be 255,0,0, but would become the duller red above since the conversion would force the more brilliant red down to the nearest point that can exist in sRGB.

Native to what?

Native in the Digital Color Meter meaning the RGB values are a reading of whatever your monitor profile is.

Apr 14, 2015 12:55 PM in response to Kurt Lang

Yes, very well explained. Thanks.


It sounds like we can trust that the Digital Color Meter is authoritative about the RGB value shown by the monitor with its current profile.


I found an interesting test file for looking at the monitor itself:

http://www.hutchcolor.com/Targets_&_images_to_go/Lab_Ramp_03.zip

(I think this is the link. Just in case, this is the Lab_Ramp_03 tiff file from www.hutchcolor.com.)


With the file open in Photoshop, its eyedropper and the Digital Color Meter show the same Lab values, which I think correspond to what the file says they should be. For a good time, open the file in Preview. On my monitor at least, the L* values seem OK but the colors do not even look uniformly gray -- they vary, some with more b* and some with higher a*. I can see the pink and green tinges across the screen. I have no idea whether this is "in spec" for a 5K Retina display but it is pretty awful.

Apr 14, 2015 1:17 PM in response to Jan Walker

When you run the cursor over any gray patch of the image, the only thing that should change is the Lightness. a and b should always be zero. With no a or b values, that means every gray step is a perfect gray with no hue of any kind. That you are seeing green and pink hues across the ramp means your monitor profile is no good.


The canned profiles that come with any monitor are only reasonably accurate for about a month, at best. And since only one panel of that type is actually profiled by someone, the rest all just get the same profile embedded in the hardware whether it's accurate to that particular panel or not. When you look in the System Preferences > Displays > Color setting, the profile listed at the top above the line is the one the OS pulls from the hardware. There will be more if you've created any with the Calibrate button.


Bad news. Trying to use Calibrate to create an accurate, or even marginally useful profile is a fool's errand. For accurate color, and a gray ramp that is actually neutral from white to black every step along the way, you need to use a real profiling solution. And eyeballing the color is not it, no matter what any software claims to be able to do without any associated profiling hardware.


Best choice for the least amount of money (in my opinion) is X-Rite's i1Display Pro. You can do about $20 better on price through Amazon.

Apr 16, 2015 12:48 PM in response to Kurt Lang

Hmm. I cannot exactly reproduce this phenomenon of the pink and green tinges.


I had made a monitor profile using a borrowed xrite i1 Display Pro and dispcalGUI. I was looking for tools to help me understand whether I used the tool correctly, which is what got me started on this thread. I guess the answer was "no" so I'll have to try again to understand how to use it.


They still look tinged but don't measure that way. This is a peculiar phenomenon that I'm seeing now though. Only the patch at L=67.87 is showing the issue. The values of a*b* reported by the Digital Color Meter vary depending on the zoom. This is in Preview. At opening, the a* and b* values are all 0. After pressing cmd-+ once, a*b* alternates between 0,-0 and 0.55,-0.41. After pressing cmd-+ again, that block shows 0.55,-0.41 across the whole thing. And, yes the iMac builtin profile shows the right values and the one I created shows the peculiar ones. Even though the iMac looks distinctly green to me. (Hey ho! off to the eye doctor!)


At this point, i'm even more mystified than I thought I was.

Apr 16, 2015 1:01 PM in response to Jan Walker

My guess would be the dispcalGUI profiling software isn't very good. If whoever you borrowed the i1Display Pro from gave you only the instrument, then you need to install X-Rite's software to use it correctly. You can get that directly from X-Rite here. Click the Software Downloads (8) heading to expand it (if it isn't already), then download the top, newest version, 1.5.6. You'll see when you run it that most of the items will say DEMO next to the various functions. But monitor profiling will be active when you plug in the i1Display Pro.

Apr 20, 2015 10:12 AM in response to Jan Walker

I've used the xrite software and wasn't very satisfied

Could you explain what issue you were having with it? I use the ColorNavigator software that came with the EIZO monitor to profile that device since it's specifically for their monitors. It sets the white point, black point and gain, then does the profiling. I have tested the X-Rite software against EIZO's, and the results are virtually identical, as far as color balance goes. X-Rite's software doesn't have a way to set the black point and gain. But then, most third party apps don't.

so moved on to the power tools.

Can't say I know what you're referring to there.

This thread has been closed by the system or the community team. You may vote for any posts you find helpful, or search the Community for additional answers.

problem with the calculator in colorsync

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple Account.